Year: 2024
Type: dissertation
Abstract: Generative artificial intelligence (GAI) poses new capabilities and ethical challenges, not least in the fields of management consulting and research. This thesis aims to explore the ethical use of GAI in this context. The topic is examined by addressing two research questions: 1) What ethical challenges and risks does the use of generative AI pose in a business management consulting and research business organization? 2) How can these risks and challenges be mitigated at the organizational level? The AI ethics literature initially addressed guidelines or principles for ethical AI and has since moved on to implementing them in practice. However, generative AI is not yet fully considered in these contexts. This thesis considers the principles as descriptions of what is (un)ethical AI and algorithmic accountability as a framework for management of ethical risks. Algorithmic accountability describes the stakeholders of an AI system and how they express accountability for it. As an element of algorithmic accountability, organizational accountability applies to the operator or developer of an AI system, including measures like AI ethics principles, AI governance measures and other implementation mechanisms for ethical conduct. As a qualitative case study with an interpretive stance and explorative approach, five interviews were conducted with the case organization’s employees. This research shed light on the ethical considerations that practitioners face when using Generative AI. The thematic analysis revealed two key themes: the ethical challenges and risks posed by GAI and the measures that could be taken to avoid these risks. The identified ethical issues partly align with the literature on AI ethics. However, information security, trust in AI and human oversight were emphasized which is not the case with Jobin’s et al. (2019) suggested consensus of five ethical guidelines. Among other issues, the participants raised concerns about data privacy, potential biases, and the integrity of AI-generated texts. As an additional finding, the lack of identified ethical risks was also brought up. In terms of avoiding the ethical risks, many ordinary measures are already taken. Despite the lack of formal measures, designated safe GAI tools are used, the use of sensitive data is careful and regulation like GDPR and the EU AI Act are complied with. Participants still expressed a desire for increasing awareness and training on ethical use of GAI. Implementing ethical guidelines or instructions for GAI use was also considered essential. The accountability regarding ethical issues was thought to eventually be on users, but before that regulators and developers should limit the possibility of negative outcomes.
Source: UTUPub (University of Turku)
Author: Isomäki, Valtteri
Institution: University of Turku
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Keywords: generative AI, AI ethics, management consulting, organizational analysis, algorithmic accountability, organizational accountability, AI governance, data privacy, algorithmic bias, information security, trust in AI, human oversight, GDPR compliance, EU AI Act, moral norms, employee perceptions
Open Access status: green
OpenAlex ID: W7065531578
